Construction of interpolating and orthonormal multigenerators and multiwavelets on the interval
In recent years, wavelets have become a very powerful tools in applied mathematics. In general, a wavelet basis is a system of functions that is generated by scaling, translating and dilating a finite set of functions, the so-called mother wavelets. Wavelets have been very successfully applied i...
|Online Access:||PDF Full Text|
No Tags, Be the first to tag this record!
|Summary:||In recent years, wavelets have become a very powerful tools in applied
mathematics. In general,
a wavelet basis is a system of functions that is generated by scaling, translating and dilating a
finite set of functions, the so-called mother wavelets. Wavelets have been very successfully
applied in image/signal analysis, e.g., for denoising and compression purposes. Another
important field of applications is the analysis and the numerical treatment of operator
equations. In particular, it has been possible to design adaptive numerical algorithms based on
wavelets for a huge class of operator equations including operators of negative order. The
success of wavelet algorithms is an ultimative consequence of the following facts:
- Weighted sequence norms of wavelet expansion coefficients are equivalent in a certain
range (depending on the regularity of the wavelets) to smoothness norms such as Besov
or Sobolev norms.
- For a wide class of operators their representation in wavelet coordinates is nearly
-The vanishing moments of wavelets remove the smooth part of a function.
These facts can,
e.g., be used to construct adaptive numerical strategies that are guaranteed to
converge with optimal order, in the sense that these algorithms realize the convergence order
of best N-term approximation schemes. The most far-reaching results have been obtained for
linear, symmetric elliptic operator equations. Generalization to nonlinear elliptic equations also
exist. However, then one is faced with a serious bottleneck: every numerical algorithm for these
equations requires the evaluation of a nonlinear functional applied to a wavelet series.
Although some very sophisticated algorithms exist, they turn out to perform quite slowly in
practice. In recent studies, it has been shown that this problem can be ameliorated by means of
so called interpolants. However, then the problem occurs that most of the known bases of
interpolants do not form stable bases in L2[a,b].
In this PhD project, we intend to provide a significant
contribution to this problem. We want to
construct new families of interpolants on domains that are not only interpolating, but also
stable in L2[a,b]or even orthogonal. Since this is hard to achieve (or maybe even impossible)
with just one generator, we worked with multigenerators and multiwavelets.|
|Physical Description:||117 Pages|